Systems Biology of human diseases

While human diseases have been traditionally studied mostly in the respective tissue they affect, it is increasingly becoming clear that most diseases are not occuring in isolation but often have systemic effects which is exemplified by abundant comorbidities (co-occurence of diseases) between diseases affecting very different organ systems. We are particularly interested in the underlying molecular mechanisms that drive the co-occurence of diseases in the context of aging and inflammatory diseases.

In the context of aging, we have recently published a comprehensive analysis of vertebrate aging on the transcriptomic level across four species and four tissues. For the first time, we were able to show that there is a core of aging-associated regulation conserved across species and tissues. Through a comparison of this core signature of aging with the signature of aging diseases we found that while transcriptomic aging is aligned with changes observed in degenerative aging diseases (cardiovascular diseases, neurodegenerative diseases and type 2 diabetes) it actually opposes changes seen in cancer. This intriguing finding is in line with epidemiological data showing a peak in cancer incidence in the 70 to 80 year olds while degenerative diseases increase in incidence up to the oldest age groups. We confirmed the existence of such an antagonism also on the genetic level. Thus, most genetic risk loci predisposing for cancer are protecting from degenerative diseases and vice versa. These observations might be one explanation for the existence of only two genetic loci associated with human longevity while there are hundreds of risk loci for aging diseases. Currently, we are focussing our work on understanding the molecular mechanisms driving the core signature of aging with a particular emphasis on the contribution of the gut microbiome.

In the context of inflammatory diseases, we are focussing on shared molecular mechanisms that underlie inflammatory diseases with extensive comorbidity between them (inflammatory bowel disease, peridontitis, COPD, psoriasis and coronary artery disease). We have identified a common core set of genes shared across these diseases and are currently characterizing it in more detail. Moreover, we use our understanding of shared processes across inflammatory diseases to develop predictors for therapy response in these diseases to biologics. We have found that therapy response can be predicted quite accurately based of specific patterns of deregulation of genes around the target of the corresponding biologic. Moreover, this information can be used to predict efficacy of therapy in other diseases and the corresponding predictions match running clinical trials very well.


Modeling microbiome x host interactions

While traditionally studied mostly in isolation, most bacteria live in complex habitats in close interaction with a large number of other organisms. Important examples are the microbial communities living on and inside the human body referred to as the microbiome. We are particularly interested in the interactions occuring in these communities, how they interact with their respective hosts, how disturbances in such communities contribute to diseases and how community composition can be modulated in a targeted manner.

To study interactions within microbial communities as well as with the host, we use constraint-based metabolic modeling approaches. Through representing microbes as well as the host with their respective metabolic networks, these approaches allow to infer the actual mechanisms underlying the interaction between microbes or with the host. Thus, these approaches have a strong advantage over the currently mostly correlative or machine learning approaches employed in microbiome research since they are not only able to infer assocations but can shed light on the molecular mechanisms underlying these associations. In this context we employ both approaches that explicitly account for the spatio-chemical properties of the respective environment (e.g. within the human gut) as well as approaches that neglect space but are computationally much less expensive (i.e. community FBA).

We are particularly interested in how the microbiome contributes to human diseases in the context of aging and inflammatory diseases. In the context of aging, we find that the aging microbiome shows very pronounced shifts in the metabolism between microbes as well as in interaction with the host. These changes are tightly linked to disease processes in the host which supports an important role of the microbiome as a driver of aging. In the context of inflammatory bowel disease, we have found particular shifts in community metabolism that are tightly linked to therapy response even prior to initiation of therapy. Similarly, we find pronounced changes in the metabolic interaction between the microbial community and its host in the early development of the microbiome in preterm infants. Intriguingly, these changes are tightly linked to the occurence of late onset sepsis which is a major cause of mortality in preterm infants. Based on these findings we are investigating how the microbiome of these infants could be modulated to prevent sepsis.

In non-vertebrate model organisms of microbiome research, we are particularly interested in the principles governing the interaction within the microbiome as well as between the microbiome and the host. Thus, in Caenorhabditis elegans we are currently reconstructing metabolic networks of bacteria living in the gut of the worm in the wild and link the metabolic capabilities of these strains to host fitness (in collaboration with Hinrich Schulenburg/Kiel University). Moreover, we study how the host adapts metabolically to colonization with different bacterial taxa. In sponges, we are building metabolic models from microbial genomes assembled from metagenomic data and relate metabolic cross-feeding patterns to spatial organization of the corresponding microbes within the sponge (collaboration with Ute Hentschel/GEOMAR Kiel).


Elucidating mechanisms of rapid microbial adaptation to changing environmental conditions

In this research area, we focus both on the optimal response to changes in conditions and the preparation for such changes in stochastic environments. Using drastically simplified models of individual pathways or entire cells, we use optimization approaches to identify regulatory programs that are optimally suited to switch between conditions.  Regarding strategies that prepare microorganisms for changes in conditions, we are particularly interested in the trade-off between growth and flexibility. This trade-off is exemplified by the two conflicting objectives inherent to microbial growth - the maximization of growth rate in a particular environment and the maximization of the ability to quickly rearrange metabolism after a change in conditions. Our research shows that both factors are important antagonistic forces that shape the allocation of cellular resources. In this area, we are closely collaborating with the group of Jakob Møller-Jensen from Syddansk Universitet (Odense, Denmark) and Jan Rupp (University Hospital Schleswig-Holstein, Lübeck) in a project in which we sample bacterial pathogens from patients with urinary tract infections in order to identify markers that indicate whether the corresponding strain is focusing on growth or flexibility. This information is of particular importance with respect to the optimal treatment strategy since strains focusing on flexibility tend to be more resistant to antibiotics than those focusing on growth.


Ewald J, Bartl M, Dandekar T and Kaleta C. Optimality principles reveal a complex interplay of intermediate toxicity and kinetic efficiency in the regulation of prokaryotic metabolism. PLoS Computational Biology., February, 2017. Vol. 13, pp e1005371. [DOI]

Development of methods for the analysis of large-scale data sets

In this research area, we apply and develop methods for the analyses of large-scale data sets. Our research is particularly focused on the utilization of genome-scale metabolic models which have been shown to be very useful in interpreting large-scale data sets. Depending on the target organism, we either use publicly available genome-scale metabolic networks or reconstruct them ourselves. Moreover, for the comparative analysis of data across several organisms, we have developed process-based methods that extend the consideration of differentially expressed genes to the detection of processes that are differentially expressed between different states. Using such methods we are able to detect common changes in expression even across very distantly related organisms since such changes are often much more conserved on a functional rather than the genetic level.